Cognate recognition
by
Silvia Aguinaga Echeverría
A thesis submitted to the Graduate Faculty of
Auburn University
in partial fulfillment of the
requirements for the Degree of
Spanish Master of Arts
Auburn, Alabama
August 4, 2012
Keywords: Second language acquisition, vocabulary recognition & processing,
cognate, phonology
Copyright 2012 by Silvia Aguianaga
Approved by
Gilda Socarrás, Chair, Associate Professor of Spanish Linguistics
Almitra Medina, Assistant Professor of Spanish Linguistics
Aimee Callander, Assistant Professor of Psychology
ii
Abstract
This study is focused on cognate recognition. First, a new classification of cognates
(English/Spanish) based on orthographic, phonological and semantic similarities between words
have been proposed-identical, similar, and partial cognates. Second, an experiment was carried
out with second semester native English-speaking of Spanish to clarify: a) whether or not
beginner level students recognize cognates in the written form by providing their English
equivalents; b) whether or not degree of similarity has an effect on students recognition and
processing of these vocabulary items. Results suggest that participants do not recognize cognates
as easily as presumed by textbook publishers and other authors. Furthermore, results from this
study show that not all cognates are recognized and processed the same. It was also indicated by
the results that a phonology overlap may play a role during recognition task.
iii
Acknowledgments
I am grateful to my parents and brother who always support me no matter what I do and a
special thank you to all of those who read it to give me their feedback. I appreciate the support
that my director, Dr. Socarrás, has offered me during this project.
Table of Contents
Abstract ......................................................................................................................................... ii
Acknowledgments ....................................................................................................................... iii
List of Abbreviations .................................................................................................................. .iv
Chapter 1 ..................................................................................................................................... 1
Chapter 2 ................................................................................................................................... 25
Chapter 3 ................................................................................................................................... 29
Chapter 4 ................................................................................................................................... 37
Chapter 5 ................................................................................................................................... 41
Chapter 6 ................................................................................................................................... 47
References ................................................................................................................................. 49
Appendix A ............................................................................................................................... 54
Appendix B ............................................................................................................................... 64
Appendix C ............................................................................................................................... 67
iv
List of Abbreviations
ANOVA Analysis of variances
BIA Bilingual Interactive Activation
BIA+ Bilingual Interactive Activation Plus
CG Control Group
CL Contrastive Linguistics
ESL English as a Second Language
FLSP Foreign Language Spanish
G1 Group 1
G2 Group 2
G3 Group 3
L1 First Language
L2 Second Language
L3 Third Language
PGs Pronunciation information guides
RHM Revised Hierarchical Model
SLA Second Language Acquisition
1
CHAPTER 1 COGNATE RECOGNITION
During the process of the acquisition of a second language (L2) in the classroom context,
a relationship between the student’s first language (L1) and the target language is assumed; this
relationship has been studied since the 1950s within the Contrastive Linguistics (CL) field (see
Di Pietro, 1971; Fisiak, 1980; Lado,1957). CL principles rest on a twofold axis, based on
behavioral theory and structuralism. On the one hand, behavioral theory predicted that
established habits of the L1 can create transference in the learning process. Transference is
defined as the use of elements from another language (usually, L1) while producing L2, e.g., the
transfer of grammar structures, e.g., native Spanish speakers while learning English as an L2
usually drop the subject pronoun a structured transfer from the L1. On the other hand,
structuralism theory strives to compare the surface of English structures to different languages.
The goal was to systematize the degrees of difference between the structure of the L1 (English in
this case) and the L2. Thus, the following equation was posited: “learning difficulties =
differences between languages structures” (Zanón, 2007). As a result, transfer theory and
analysis of errors emerged as a new linguistic perspective. However, research does not support
transfer as the main source of error production in second language acquisition (SLA). The
majority of studies focused on error analysis concluded that around only 30 percent of errors can
be attributed to negative interference effects (e.g., Dulay & Burt, 1973; Ellis, 1986).
Nevertheless, although the popularity of CL analysis decreased, especially after generative ideas
appeared (Chomsky, 1959), studies have been conducted within the phonology field addressing
the difficulties usually learners have due to their native language (see González Barrera, 2010;
Moreno Fernández, 2000; Serradilla Castaño, 2000; Terrel, 1989).
2
Vocabulary acquisition plays a central role in the acquisition of a second language, as
pointed out by Gass (1990), Levenston (1979), Oxford and Scarcella (1994), Politzer (1978),
among many others. In particular, vocabulary acquisition is essential not only in communication,
but also in understanding the target language (e.g., Levenston, 1979; Politzer, 1978).
Nonetheless, as Lado (1955) indicates, target language vocabulary is not the only element to
consider:
We simply cannot ignore the native language of the student as a factor of primary
importance in vocabulary, just as we cannot ignore it in pronunciation and
grammatical structure…Similarity and difference to the native language in form,
meaning and distribution will result in ease or difficulty in acquiring the
vocabulary of a foreign language. (p. 31-32)
The focus of the present study is on the role cognates (i.e., words with similar
morphology, sounds or meanings in two languages) play in the acquisition of L2. In particular,
this study addresses the recognition and processing involved when learners encounter these
vocabulary items. Cognates have been identified as helpful and easy vocabulary items to learn in
a second language (De Groot, Dannenburg, & Hell, 1994). However, only a few studies with
novice language students have been done, with most of them assuming that beginners can
recognize cognates, and as a result take advantage of the similarities between these vocabulary
items in the acquisition process. Moreover, most language textbook publishing companies also
assume that learners can recognize and process cognates without problems; for this reason
practice activities on cognates recognition are scarce in those publications. The present research
intends to address this topic, that is: What is the role of cognates in SLA? Specifically, are
cognates between English/Spanish recognized and processed by Spanish L2 novice learners? If
3
so, are all cognates processed equally by learners or are there differences pertaining to the degree
of morphological and phonetic similarities perceived by students? For example, are there
processing differences between identical cognates such as “mural”–“mural”, similar cognates,
i.e., those which differ by one or two consecutive letters, such as “inherent”–“inherente”, and
partial cognates, those which differ by two or more letters, such as “conclude”–“concluir”?
These cognates’ characteristics will be defined later in Chapter 3.
Theoretical Foundation
Many studies have focused on how learners store and process non-native vocabulary
(e.g., Chen, 1990; De Groot et al., 1994; Kroll & Cureley, 1988; Potter, So, Von Eckardt,
Feldman, 1984), but in particular, three memory and processing models have been put forth in
the literature to address this topic: the word-association model, the concept-association mediation
model (both studied by Potter et el., 1984; Chen, 1990; De Groot et al., 1994), and the
intermediate model, later called revised hierarchical model (Kroll, & Cureley, 1988; Chen &
Leung, 1989; Kroll & Stewart, 1994). These three models share two major assumptions
regarding vocabulary acquisition and production. First, they assume the existence of two separate
and independent lexicons for each language, in this case the first language (L1) and second
language (L2). Second, they support the assumption that these lexicons share a similar concept
representation for each word. They disagree, however, on the proposed connection between
memory stores, i.e., where the information is stored, and the manner in which new words are
processed (see Appendix-B Figure1 from De Groot et. al, 1994).
The word-association model postulates direct connections between L2 and L1 lexicons
and each L2 word is connected to the concept through its equivalent L1 word (see Figure 1 in
Figure 1). As a result, L2 words can only access the corresponding concept indirectly via L1. In
4
contrast, the concept-association mediation model proposes that both languages operate
independently and they both are connected to one another only through the concept (see Figure 2
in Figure 1). The third model—the intermediate model or revised hierarchical model—combines
the two previous models and assumes that both type of access, via concept or via L1, occur but
they depend upon the type of task or the proficiency level of the student (see Figure 3 in Figure
1).
Potter et al. (1984) compared the word-association model and the concept-association
mediation model in a study in which proficient Chinese-English bilinguals and novice English-
French bilinguals performed picture and word naming tasks in both directions between L1 and
L2 translation. The authors also studied latency during reading aloud protocol, word-translation
tasks, and picture-naming task. Latency is defined as the time that elapses between the stimulus
(the target word or the picture) and the student’s response. On the one hand, the researchers
predicted that subjects should spend less time on the word-translation task (translating words
from L1 to L2) than on the picture-naming L2 task, based on the word-association model
students will have to complete more steps in the picture-naming task, i.e., participants have to
activate the concept. On the other hand, Potter et al. predicted, in line with the conceptual-
association mediation model equal response times for the two tasks were expected based on the
assumption that pictures and words share a similar amount of time for semantic-access (see
Appendix-A Table 1 from Potter et al., 1984).
The results of Potter et al. support the concept–association mediation model in that
participants spent the same amount of time on both tasks, translating and picture-naming.
Similarly, the authors found evidence in support of direct access between the two lexicons; they
point out that “naming in a second language can be cued directly by pictures that have never
5
been associated with the L1 word. These relationships are embodied in the concept mediation
model” (pg. 36).
The third model, the intermediate model or revised hierarchical model (RHM), has been
proposed by researchers such as Kroll and Cureley (1988), Chen and Leung (1989), and Kroll
and Stewart (1994), and RHM was presented as an alternative model of word-association and
concept-association mediation models. They argue that in order to learn an L2, it is necessary to
use an existing internal representation. Kroll and Stewart (1994) focused their research on
asymmetries in translation performance between backward and forward translation. Their study
was carried out with word lists were presented in semantically categorized (all words belonging
to the same semantic category, e.g., all kitchen utilities) or in mixed lists (words selected from
several semantic categories and presented in a random order). They found that the semantic
contexts affected forward translation, but not backward translation. Another finding was that the
translation of words in categorized lists took longer than the translation of words in mixed lists.
That is, while translating from L1 to L2, translation would be semantically mediated because of
the strong L1 meaning link. However, translation from L2 to L1 could be accomplished
lexically, without semantic access, if the L2 word enabled lexically mediated retrieval of the
translation.
Pertaining to cognates, Sunderman and Kroll (2006) compared the performance of less
and more proficient L2 learners on a translation recognition task. Both groups had to decide if
two words were translation equivalents of each other. The stimuli were presented so that the L2
(Spanish) word appeared first, followed by the L1 (English) word. The results showed that only
learners at early stages of L2 acquisition indicated evidence for activating the L1 translation
equivalent; lexical decisions were faster in cognates than in non-cognates. That is, the authors
6
noticed a data pattern supporting the word-association model for low-proficiency bilinguals and
the concept-association mediation model for high-proficiency bilinguals. These results suggested
that learners, for whom the L2 is relatively weak, will exploit the L1 translation equivalent for
the purpose of accessing meaning. Researchers also pointed out that bilinguals are at an
advantage in recognizing words that are cognates, whereas monolinguals show none of these
effects.
Another branch of research that complements the field of L2 word recognition has been
explored by Ton Dijkstra and Walter Van Heuven (1998). Those authors have worked on
bilingual word recognition and developed initially the bilingual interactive activation model
(BIA) and later on, the bilingual interactive activation model plus (BIA+). The BIA model is
based on the interactive activation model first proposed by McClelland and Rumelhart (1981).
BIA model supports the non-selective nature of lexical access across languages. The bilingual
lexicon is integrated and the lexical access is non-selective; lexical candidates in both languages
are active whenever the input shares features with alternatives in each of the languages. Opposite
to RHM, which suggests that the manifestation of L1 activation during L2 processing consists of
direct access to the L1 translated equivalent, Dijkstra and Van Heuven suggested that it is not the
translated equivalent itself that is activated, but rather lexical sound relatives. The authors
conducted a study with fluent bilingual Dutch-English speakers and found out that when subjects
performed lexical decisions in L2, words in both languages L1 and L2 that are orthographic
neighbors of L2 target word, have an influence on their performance. The model consists of a
networking of nodes representing orthographic, phonological and semantic representations. In
this model the main function of the language nodes is to determine which lexicon to inhibit. For
7
instance, upon seeing an L2 word, a native English speaker’s nodes will inhibit words similar to
his/her L1 (see Appendix B, Figure 2 from Dijkstra et al. 1998).
Dijkstra and Van Heuven have pointed out in several experiments that the activation of
lexical form information of L1 to L2 cannot be only applied to the processing of cognates and
interlingual homographs (words that share form but not meaning across languages), but also
extends to any words that share orthography-forms; e.g., “fort”-“kort” (translates to “brief” in
English). Also see Dijkstra, Van Jaarrsveld & Brinke (1998); Dijstra & Van Hueven (2002);
Dijstra & Van Hueven (2002).
As proposed in the BIA model, when a proficient bilingual reads a letter string (a word),
several lexical candidates—regardless the language—are activated. In addition, these effects
appear to be driven in a bottom-up manner (letters activate words from both languages in an
integrated lexicon) so that they are relatively uninfluenced by factors such as instruction or the
language chosen for the task. Also, the language nodes exert a top-down processing (language
nodes selectively inhibit effect in words of the other language).
The authors bring up a very interesting point, that is, the existence of a phonological
effect during a task recognition (Dijkstra, Grainger & Van Hueven, 1999). This new point, not
taken into account by the aforementioned studies, provides a very interesting and new field to
research. The findings showed cross-language effects of phonological overlap between words;
while orthographic and semantic overlap were shown to result in facilitatory effects relative to
controls, phonological overlap induced inhibition. In sum, they pointed out that identical
cognates and interlingual homographs can be identified faster because they share lexical and
sublexical orthographic representation across languages. In contrast, the study showed a negative
influence of cross-linguistic phonological similarities on word recognition latencies in bilinguals.
8
As the authors mentioned: “phonological inhibition now occurs because after a given letter string
activates all compatible phonological codes independent of language, this competition results in
a delayed identification of the item in the target language” (p. 512).
The phonological interference and other issues found through several experiments—such
as the relationship between word identification and task demands, the representation of
interlingual homographs and cognates, and representational and functional aspects with respect
to the language nodes—generated a new diagram to represent the model: BIA+ (see Appendix B,
Figure 3 from Dijstra & Van Heuven, 2002). The improved model rests on the following ideas:
a) Bilingual word recognition is affected not only by cross-linguistic orthographic similarity
effects, but also by cross-linguistic phonological and semantic overlap. b) The first stages of
word recognition are carried out in the same manner as in BIA model; several lexical
candidates—regardless the language—are activated depending on their similarity to the input
word, and on other individual factors such as use frequency, subjective frequency, L2
proficiency, etc. c) Orthographic representation becomes activated at the same level as
phonological and semantic representations. As activations depend on individual factors, this
implies that the phonology and semantic L2 codes may be delayed in their activation in relation
to the L1 codes; the authors relate this phenomenon to the type of task and the language level of
students. d) With linguistic or non-linguistic context effects, BIA+ model predicts that the type
of task will have an influence on word recognition processing. When a word recognition task is
inserted into a sentence context, the process is sensitive to syntactic and semantic context
information; for example, context information might inhibit or reduce the activation of lexical
candidates or induce a more flexible activation of lexical candidates in the two languages.
9
Empirical Research on Cognates
The studies reviewed for the present research point to the fact that even though a precise
definition of cognates is warranted in order to fine tune research objectives in this topic, no
consensus is found in the literature pertaining to their definition, as pointed out by Friel (2001).
Also as Friel (2001) notes, most researchers agree that cognates are words with similar roots,
hence their similarities in sound and appearance. Hall (2002) defines them as “words in two or
more languages which share phonological and/or orthographic forms, and normally -but not
necessarily- are also related semantically” (p. 69). Holmes and Guerra (1993) defined them as
items of vocabulary in two languages that have the same roots and can be recognized as such. It
has been observed that even though researchers agree on three fundamental properties of
cognates, namely their phonetic, orthographic, or morphological and semantic similarities in the
languages compared, they do not clarify the meaning of similarity.
Lobo (1966) created a cognate corpus of 10,000 words shared between Spanish and
English. Lobo categorized cognates as follows:
1) A cognate is considered to be an item with a shared orthography, meaning, or etymology,
or a combination of these overlapping in two languages.
2) Three classes of English-Spanish cognates were distinguished:
a) Those which are similar in orthography, meaning, and etymology, called true
cognates. (For example: “tomato” –“tomate”)
b) Those which are similar in meaning and orthography but not in etymology are
called accidental cognates. (“Shock” – “chocar”)
c) Those which are similar in orthography and etymology, but not in meaning are
false cognates. (“Bigot” – “bigote”)
10
In addition to Lobo’s categorization, other researchers have proposed different methods
to identify cognates and create different cognate corpora. Friel and Kenninson (2001) researched
two techniques to identify cognates and as a result were able to create a list of English-German
cognates. They worked with 250 undergraduate students, native speakers of American English
with no previous learning experience with German. They used two techniques to identify
cognates: Technique 1 included the similarity-rating procedure of De Groot and Nas (1991).
During this task participants were asked to rate the similarity of translation pairs on a seven-point
scale (1 = “Low similarity” e.g., “sarg”- “coffin” and 7 = “High similarity” e.g., “kanal”-
“canal”). Technique 2 included the modified translation–elicitation task of Kroll and Stewart
(1994), which required students to translate a series of German nouns, as “kompromiβ”-
“comprimise”. With both techniques, half of participants were instructed to base their rating on
the overlap in sound and appearance between translation pairs where they could hear the
pronunciation of the word. Conversely, the other half rated the pairs and performed the
translation task based only on appearance. As a result, they obtained a list with 112 German-
English cognates and 94 false cognates. A correlational analysis between Technique 1 and 2
indicates that both were significantly positively correlated. Therefore, they obtained the same
cognate translation and rating results with both methods.
Furthermore, four hypotheses were tested during their experiment. Hypothesis 1 assumed
that participants would translate nouns with umlauted vowels and/or asszets (special German
characters) less accurately than other nouns because the orthographic differences would interfere
with word recognition. The results showed that the orthographic differences had a significant
effect on both tasks with considerably lower results for those pairs in which the German word
contained a German-specific character. However, this conclusion has to be examined more
11
carefully because there were more words with specific German characters than words without
these characters, as pointed out by the researchers. Also, they do not mention if they kept the
number of the specific characters as a constant in each word, nor they specify how many
characters were in each word. This could be an intervening variable.
Hypotheses 2 and 3 postulated that participants who received pronunciation information
guides (PGs) would produce higher similarity ratings during Technique 1 because the sound
would help students recognize similarities between languages. In the same manner, participants
would rate German-English translation pairs as more similar than those who did not receive PGs.
As a result, these researchers showed that performance in the similarity-task and translation-task
was influenced by the PGs, such that they (the PGs) had a positive influence on students’ ability
to recognize cognates. The differences between the +PGs and –PGs groups, however, were not
significant.
Hypothesis 4 predicted an interaction between the two factors (PGs and orthographic
differences) in both techniques; that is, the researchers expected the participants who received
the PGs to perform better at identifying similar word pairs than other participants when
encountering German words with no English characters. However, the influence of the PGs was
different between both techniques pertaining to those words that contained German characters.
Specifically, they saw that for those words that did not contain German characters, the PGs had
more influence in the similarity-rating task than in the translation-elicitation task.
It is important to point out that the similarity-rating method seems very ambiguous. To
scale 563 words with a rating from 1 to 7 can prove to be a difficult, tediousness and confusing
task. Furthermore, the translation-task might have proven more difficult because the cognitive
12
demands look to be extensive (asking participants to translate a word that he/she has never
encountered) even with the help of the PGs.
The studies included in this literature review do not share similar results. Moreover, they
can be characterized as lacking well-defined conclusions and often presenting conflicting results.
As well, even though most of the studies work with bilinguals subjects, there is no clear
definition of what bilingual means. This fact makes their findings more difficult to understand.
Holmes and Guerra (1993), Tonzar, Lotto and Job (2009), Hall (2002, 2009), Tercedor
(2010), Hoshino and Kroll (2008), Costa, Caramazza and Sebastian-Galles (2000), and Sherkina-
Lieber (2004) show that students can recognize cognates, and due to this fact they can easily
process these words. On the other hand, Lightbown and Libben (1984), Tréville (1996), and
Harley, Hart and Lapkin (1986) conclude that students do not recognize cognates and due to this
fact, instructors should teach students how to recognize and work with cognates., The studies
discussed point to need for further clarification of the role that cognates play in SLA, one of the
focuses of the present research.
An initial description of several studies that suggest that students do recognize cognates
will be presented. Additional studies will later be described which found different results
showing that students do not recognize cognates.
Holmes and Guerra (1993) proposed a method to recognize cognates in reading
comprehension tasks. The researchers hypothesized that students can recognize cognates easily
because of the word similarities. The authors worked with undergraduate Brazilian students of
Portuguese Philology and graduate Brazilian students of Cognitive Psychology. Most of these
students were almost complete beginners in learning English with only a vague knowledge of a
few words from previous school experience. Data was gathered through Think-aloud protocols
13
(TA): students read silently through an English text and orally summarized their understanding
in Portuguese. At the same time, they spontaneously commented on the task. The analysis was
done by matching the wording of the subjects’ summaries with the original text to identify
cognates and other vocabulary items which were recognized or misrecognized in the original text
and were transferred directly to the summary. In conclusion, they considered cognate recognition
as a “natural” strategy; however, there was a variation of recognition from one student to
another. To explain these variations, the authors stated that “cognate identification seemed to be
personal, with some subjects inclined to be more liberal than others admitting a word cognate”
(p. 89). Holmes and Guerra also pointed to the idea that previous knowledge was important.
Students demonstrated that they were very dependent on their previous knowledge of the topic to
predict the text’s content and the author’s purpose. Students also used their previous knowledge
to interpret cognates.
Another technique that was used by the authors is cognate recognition in group summary
preparation, which focuses on false cognates. The chosen task was to write in groups a summary
in Portuguese (the students’ L1) of a given English text using “directed summary” (p. 94).
Authors decided to carry out the activity in groups because it was an activity that students
usually do in a classroom environment. As a result, they discovered that almost all cognates were
recognized.
Based on these results and on the demonstrated importance of previous knowledge, it can
be noticed that students of Portuguese have knowledge of word forming. Due to this, these
results could also indicate that for these students it was easier to recognize cognates because of
their linguistic knowledge. Also, a limitation in the group work should be considered: Were
cognates universally recognized or just by the best students in each group? However, the authors
14
do not mention it. In addition, another interpretation to be considered from reading this article is
that students are not always “liberal” in recognizing cognates because, perhaps, they do not
recognize them.
An additional study within a classroom environment was done by Tonzar et al. (2009),
who worked with Italian native speakers. The authors compared two teaching/learning methods
(word–word, picture–word) and evaluated vocabulary acquisition with different age groups. In
addition, Tonzar et al. tried to assess the role of cognates’ status in the learning process.
In this study, participants were Italians in Grades 4 and 8 studying English and German
as their L2. Fourth grade students did not have previous L2 language knowledge; eighth grade
students had four years working with an L2 in school. The division of the groups was decided by
their age: Group 1 was formed by 123 fourth graders, approximately 9 years old, and Group 2
was formed by 106 eighth graders, approximately 13 years old. Students were evaluated using a
test, which will be explained later, administered in four different sessions. They compared a
verbal method (L2 word + word in the native language) and a picture-based method (L2 word +
picture). As their materials, the authors used 40 cognates within their pictures list.
In Phase 1 students did two pretests; the first test was a picture-naming test in their L1,
and the second test was a cognate rating-task to assess the cognates’ relationship between Italian
words and their translation in English and German. This rating-task was like Friel and
Kenninson’s (2001), whereby students evaluated the cognates using a 7-point scale of
orthographic similarity. As a result of the pretests, authors obtained a cognate list for the actual
experiment—a list of 40 pictures with the corresponding names in Italian, English, and German.
In phase 2, Tonzar et al. tried to single out the effect of the learning method in relation to
the linguistic distinction between cognates and non-cognates. They showed each item from the
15
list of words for each of the two different methods (verbal method vs. picture-based method).
Finally, students took four exams: the first one was after the first session; the second one was
after the second session; the third exam was one week later; the fourth exam was one month
later. The third and the fourth exams were unannounced.
The results were the same in both grades: participants performed better with the picture-
naming method than with the word-word method. Also, they were better able to remember
cognates than non-cognate words after several weeks, showing that cognates are easier to learn
than non-cognates.
After reading the article, it should be noted that cognates help in the learning process
even at the beginner level; however, do students really recognize or notice them? Can the fact
that students saw the pictures in the first test and named them in their native language have an
influence on the results?
Other studies on the role of cognates in learning an L2 or L3 have also been conducted by
Hall (2002, 2009). Based on his parasitic model of vocabulary development, he assumed that
learners automatically utilize material from L1 or L2 in order to establish an initial representation
of the unknown term; this implies that learners are predisposed to produce overlap between
languages. On encountering novel vocabulary items, learners initially use already existing
information from the words that they know in order to confirm or create hypotheses about the
meaning of these novel words. Learners, in turn, activate lexicon which are based upon the
connectionist view that the mind is a vast network of simple processing units where complex
mental states and behaviors are the results of different configurations of the network. Hall
(2002), worked with 95 university students with an intermediate English level, and with Spanish
as a native language. He used a word list of 100 items--10 real English cognates and 90 words
16
divided in different sets: one set of 30 English non-words was constructed; half of them looked
like Spanish cognates (pseudo-cognates, such as “campanary”= “campanario”); the other half
(another 15 words) did not look like Spanish cognates (“pirt”=“estribo”). The remaining words
(60 real English words) were composed of half cognates and half non-cognates. These 10 real
English items were used as warm-up to explain the experiment process to the students.
Participants were asked to perform two tasks for each type of word they saw--pseudo-
cognates (non-word cognates and non-word non cognates) and real words. First, they had to
record whether or not they had seen the word before; secondly, they had to write down what they
thought the Spanish word closest in meaning to the English word presented could be, even if
they had to guess.
Hall based his classification of pseudo-cognates and real words on four criteria
(frequency, length, morphological complexity, and cognate status--real words and non-words
must share at least two-thirds of their form) designed to address five separate hypotheses. Results
showed that, first, there were a high number of students that recognized familiarity with the
pseudo-cognates, and less familiarity with the other two types of words: non-word non cognates
and real words. Second, the number of different Spanish translations given per item was lower
with the non-word cognates group within the category of pseudo-cognates. Third, the number of
participants responding with the most favored translation per item was higher for pseudo-
cognates. Fourth, the number of participants answering with forms that shared the initial letter in
Spanish and in the pseudo-cognates were also higher. And fifth, that the number of participants
responding better on forms that shared three consonants between Spanish and pseudo-cognates
was higher. Hall concluded that an overlap between languages exists.
17
As can be noted this experiment lacks context, which has also been seen with various
other experiments within this literature review where only a list of words has been given to the
participants. Could students’ reactions be different if cognates were framed in a particular
context? Notwithstanding, the main limitation found within Hall’s study has been the list of
pseudo-cognates because these pairs are very similar (“campanary” = “campanario”, “stribe” =
“estribo”). Could students recognize words with less similarity? It is also unclear as to why Hall
used the thirty real cognates in his experiment, and what the results were regarding those words
in particular.
In a recent study, Tercedor (2010) worked with cognates but from a different perspective.
She wanted to describe the role of cognates as lexical choices in translation, in contrast with
previous pedagogical approaches that have been applied in previously discussed investigations.
Within Tercedor’s article, two groups are compared: 77 Spanish university students studying for
a Translation and Interpretation degree at the University of Granada, and 66 volunteer students at
the University of Ottawa who were considered to be advanced learners of Spanish. Both groups
did a translation task that differed in format. The first group, the Spanish university students
from Granada, received a diagram of an online business webpage in .html format. Participants
were instructed to translate from English to Spanish using a Computer Aided Translation tool
(CAT) or text editor, and they were given credit for their work. Students submitted the
assignment electronically within one week. The second group, the Advanced Spanish students
from Canada, viewed an onscreen presentation of authentic English sentences in context with the
same vocabulary that the other group had seen. These contexts contained the same cognates as
part of compound structures or phonological units in an equibiased mode, i.e., with two or more
translation equivalents not biased in frequency.
18
The Spanish translation students did not use cognates or reduce their use of cognates
when translating because, for them, cognates are considered to be potential translation problems.
However, Advanced Spanish learners translated cognates more often.
It should be noted from this article that the Spanish university students did not use
cognates because in those specific sentences the cognate translation did not match the context.
As Tercedor pointed out, cognates can be a potential translation problem due to the fact that they
can be inappropriate for translation purposes because not all of them match every context.
Similarly to these findings on the importance of the word-context, Lightbown and Libben (1984)
also concluded that not all cognates can be translated in all sentence contexts. However,
Advanced Spanish learners seemed to use cognates without fear, presumably due to the fact they
do not find them to be problematic in most contexts. Due to this, they used cognates
inappropriately in some translations because the outcomes of the translations did not work in
some contexts.
In gathering research done on cognates, very few experimental studies have been
encountered within the field of applied linguistics pertaining to the recognition and the effects of
cognates on vocabulary learning. Furthermore, it has become obvious that most of these research
studies have been focused primarily on bilinguals. For example, Hoshino and Kroll (2008)
completed a study within the field of psycholinguistics that focused more on processing
vocabulary, in which they paid particular attention to the value that cognates had in studying
lexical access and representation in the bilingual lexicon. Their research focused on picture-
naming, and they worked with 35 Spanish-English and 20 Japanese-English bilinguals with an
L1 dominant. These subjects were placed in front of a computer and asked to name the pictured
objects in English (with the written lexical form being absent) as quickly and accurately as
19
possible. If they did not know the name, they were asked to say “no” into a microphone and their
answers were recorded throughout the study. A total of 20 practice trials took place during the
experiment. The results revealed that one important effect of using cognates was that the students
were able to name pictures faster when the word was a cognate. Moreover, the pattern of results
was similar for both groups (Spanish and Japanese), suggesting that even when a bilingual’s two
languages do not share script, there is cross-language activation; in this case the phonetics of the
non-target language were utilized.
In another study, Costa et al. (2000) worked with Catalan–Spanish bilinguals (they
reported that Catalan was their dominant language), who were asked to name two sets of pictures
whose names were or were not cognates in the L2. The participants performed two tasks. In the
first task participants named pictures in their dominant language (Catalan); the authors executed
the same task with monolingual Spanish speakers to compare the naming latencies of both kinds
of participants. Within the second task, the authors worked only with the Catalan speakers who
named pictures in both languages, Catalan and Spanish.
They based their work on the cascaded activation assumption that the phonological
content of the word is activated through the semantic system, and this phonological activation
flows continuously from the lexical layer to the phonological layer, regardless of which one is
selected. Based on this theory, the authors showed 80 pictures (40 cognates, 40 non-cognates) in
a manipulated order based on the syllable length, and the frequency of the picture names. All
cognates shared at least the first whole syllable, and all of them shared at least the first phoneme.
Costa et al. hypothesized that if the two sets of pictures were comparable with respect to
variables that affect naming latencies, Spanish monolinguals would show identical latencies for
the two sets of pictures. Furthermore, if cognates affect picture-naming latencies, a difference
20
between the two sets of pictures would be produced by the bilinguals. The study showed that
bilinguals and monolinguals produced very different patterns regarding the cognate variable.
Bilinguals named the pictures whose reference was a cognate faster than pictures with non-
cognates; on the other hand, monolinguals named the two sets of pictures equally quickly. A
second task was carried out to investigate how cognates affect the performance of bilinguals.
Within this task, half of the participants reported that Catalan was their dominant language and
the other half that Spanish was their dominant language. Costa et al. showed the same 80 pictures
that were shown in Experiment 1, and as a result the main effect of block repetition was
significant in the naming latencies analyses. In summary, results suggest that cognates facilitate
the naming performance of bilinguals.
To obtain more information about bilinguals and the hypothesis of sharing a lexical
representation between both languages, Sherkina-Lieber (2004) worked with 40 Russian-English
bilinguals and 20 English monolinguals. She performed a 10-point scale rating-task to evaluate
the frequency of use for a set of words, 35 cognates and 55 non-cognates. Bilinguals rated the
frequency of cognates significantly higher than monolinguals did. However, there were no
differences for non-cognates. Due to this fact, the author concluded that cognates have an effect
in frequency rating, and this cognate frequency effect can be possible if cognates share at least
some part of their representation between the two lexicons.
In contrast, several studies have concluded that students do not recognize cognates, and
due to this fact instructors should include more cognate recognition activities in their classes
(Lightbown and Libben (1984), Tréville (1996), and Harley et al. (1986)).
Lightbown and Libben’s (1984) research goal was to explore the role of transfer in the L2
lexicon, or more specifically, the use of cognates. The research done was based on comparisons
21
of free compositions, a cloze test (consisting of a portion of text with certain words removed,
where participants were asked to replace the missing words), and a word acceptability judgment
task performed by two separate groups—one consisting of native French speakers learning
English as an L2 (divided into two groups) and another consisting of native English speakers.
A pre-test was administered to native French speakers to identify how much contact
students had with English outside of their classes. Afterwards, the task consisted of a movie
which was shown twice to each group; then students had approximately 40 minutes to write a
composition based on the movie. They were instructed to write about the movie as if they were
telling the story to someone who had not watched it. Four months later, a portion of these ESL
students also wrote another composition about the same movie in their native language, French.
The authors considered 10 cognates as potential sources of overlapping, such as cowboy,
saloon, sheriff, or guitar. Their results were inconclusive because their Francophone participants
did not use the same words in the same contexts as did English native speakers. To address this
issue, the researchers carried out another experiment in order to judge the appropriateness of the
words, and the appropriate context with which to study the cognates that they chose as a
potential source of overlap. This new experiment only differed by one step from the original
one—a test with 10 concepts that was administered after watching the movie. This test tried to
“force” students to rely on cognates more heavily; however, it did not succeed. The authors
created a cognate classification criterion related with the context after seeing that not every
student used the cognate in every context:
1. Appropriate cognate in both languages.
2. More appropriate in French
3. More appropriate in English
22
4. Appropriate in neither language
Also, authors mentioned that it is possible students do not trust words with similar spellings in
the two languages or that students simply are unaware of the relationships between two
languages. Due to this fact, students need to be taught how to recognize all the potential
relationships between two languages.
Support for Lightbown and Libben’s findings on the importance of training students to
recognize cognates is found in Tréville’s (1996) study on lexical reading and cognate recognition
in French as an L2. She worked with 105 university level beginner and false beginner learners of
the L2, all of whom were Anglophone. Students were registered in a course focused on French
listening and reading comprehension. Pupils were divided into two groups--the experimental
group participated in a modified portion of the course focused on cognates, while the control
group did not receive this portion of the course. At the end of the course, all students
(experimental and control group) were given a special exam, the TARC (Test of Aptitude in
Recognizing Written Cognates). The exam involved s 7 scales, two of which were separated into
two subdivisions. The exam in its entirety consisted of a total of 9 tasks, including:
1) The recognition of cognates and French grammatical inflections out of context.
2) The transfer and application of intertextual correspondence rules.
3) The identification of grammatical categories of English (Task 1) and French
words out of context (Task 2).
4) The identification of grammatical categories of English (Task 1) and French
words in context (Task 2)
5) Derivation based on French words.
23
6) Comprehension of a French text of 180 words, including a large number of
cognates.
7) Inserting cognates in the correct form in appropriate contexts.
Tréville found that the experimental group obtained better results in recognition,
application of interlexical correspondence rules, identification of grammatical categories,
generalization of interlexical rules, and the selection of appropriate words in given contexts.
It should be noticed, Tréville’s study points to the fact that teachers or textbook
publishers should not presume that existence of cognates between languages will guarantee that
L2 learners will use cognates properly.
Harley et al. (1986) presented an English-French bilingual study with Americans living in
Canada, and to my knowledge there are no other studies related to this idea.
As previously mentioned, Harley et al. (1986) reached the same conclusion in their study
as Tréville 91996), and Lightbown and Libben (1984): “one cannot assume that the existence of
cognates between languages will ensure that L2 learners will, without instruction, use or even
recognize all the potential relationships between languages” (p. 407). The study done by Harley
et al. focused on bilingual children in a French L2 immersion situation. Their hypothesis
proposed that early bilingual schooling would enhance their performance on various kinds of L1
tasks, and as a variable they also studied the role of cognates. They carried out a longitudinal
study with participants from grades 1 to 6. To approximate the level of their L1-English,
participants took the Canadian Test of Basic Skills (CTBS). The results from the vocabulary
section showed that immersion students performed better than regular program students. The
authors wanted to determine whether or not knowledge of cognates in French was an advantage
for immersion students. That is, whether or not immersion students would perform better than
24
regular students overall on a test of vocabulary knowledge test and with English lexical items
that were cognates in French. This hypothesis was not supported by the results of this test. In
relation to these findings, the authors concluded that students do not use these relationships
between languages.
As has been pointed out before, the studies covered in this literature review do not share
similar results. As well, most of the articles can be described as lacking of well-defined
conclusions and often presenting limitations.
25
CHAPTER 2 RESEARCH QUESTIONS AND HYPOTHESES
The aim of this study is to define the role cognate recognition and processing plays in the
acquisition of Spanish as an L2 vocabulary. Research focus on this topic, is scarce and what has
been discussed in this review represents, to my knowledge, the most relevant studies that have
been done on cognate recognition, processing, and translation. Tercedor’s (2010) research
focused on translation, and Harley et el., (1986), Hoshino and Kroll, (2008), Costa et al., (2000),
and Sherkina-Lieber, (2004) focused their studies on bilinguals, however, the definition of
bilingual was not always clear. Studies conducted by Lightbown and Libben (1984), Tréville
(1996), Holmes and Guerra (1993), Tonzar et al., (2009), and Hall (2002, 2009) worked with
novice levels.
The present study attempts to address the following research and methodological issues.
First, provide a more precise definition of what cognates are in order to study their role in SLA.
Second, clarify what is meant by the similarities they share between languages (i.e., English and
Spanish in this case), to address the methodological issues discussed in Chapter 1. Third, shed
some light on the conflicting results available on cognate recognition. As mentioned in Chapter
1, on the one hand, studies presented by Lightbown and Libben (1984), Tréville (1996), and
Harley et al., (1986) conclude that students have difficulties recognizing cognates. On the other
hand, Holmes and Guerra (1993), Tonzar et al., (2009), Hall (2002, 2009), Tercedor (2010),
Hoshino and Kroll (2008), Costa et al., (2000), and Sherkina-Lieber (2004) pointed out that
students can indeed recognize cognates as a natural strategy for understanding language.
Given these gaps in L2 research and the impact they can have on vocabulary acquisition,
examination into the following questions are both timely and critical.
26
Research Questions:
The following research questions will guide the present study:
1. Do second semester native English-speaking students of Spanish as an L2 recognize
cognates in the written form as it has been assumed by beginner Spanish textbook
publishers? And if they do, what role does the L1 play in the processing of cognates?
2. Does cognate degree of similarity have an effect on student recognition and processing
of these vocabulary items? That is, does a classification of cognates predict their
performance?
3. Does the use of cognates facilitate reading comprehension?
Hypotheses
The present study addresses several hypotheses. First, in terms of cognate recognition,
whether English-speaking students of Spanish as an L2 are able to recognize all the cognates
displayed within the context of a text, as put forth in the literature in studies by Holmes and
Guerra (1993), Tonzar et al., (2009), and Hall (2002, 2009) who proposed that cognates are easy
words to process due to the similarities; or whether students are not able to recognize cognates,
as proposed by Lightbown and Libben (1984), Harley et al. (1986), and Tréville (1996). These
studies predict that second semester English-speaking students of Spanish as an L2 will not
recognize all the cognates displayed within the context unless instructors teach students how to
recognize and work with cognates. At the same time, these hypotheses are related to the effect
that cognates can have on comprehension and processing.
As proposed by several authors similarities between words can have an effect in cognate
recognition. Since one of the main goals of the present research is the definition of cognate
similarity between English and Spanish, cognates were classified into three different levels of
27
similarity: identical, similar, and partial (see Chapter 3 for details). I predict that most of the
cognates that will be recognized by the participants will be identical or similar cognates because
of the greater similarities between the English/Spanish forms of the words.
Several models have been proposed in the literature to account for cognate processing
and production in language acquisition, e.g., the word-association model, the concept-association
mediation model (both studied by Potter et el., 1984; Chen, 1990; De Groot et al., 1994), the
revised hierarchical model (Kroll, & Cureley, 1988; Chen & Leung, 1989; Kroll & Stewart,
1994), and the bilingual interactive activation model plus (Ton Dijkstra & Walter Van Heuven,
1998), see Chapter 1 for details. For the present study, two of the models reviewed are relevant:
the RHM and the BIA+ models. First, the RHM proposes that vocabulary can be processed in
many ways, i.e., via the L1 for novice bilinguals or via the concept for more advanced bilinguals.
Notice that the RHM has been presented in the literature as a production model as stated by
Kroll, van Hell, Tokowicz, and Green (2010): RHM “was not primarily a model of word
recognition but a model of word production” (p. 2). However, RHM has been taken into account
in the present study because it is assumed that learners need to process vocabulary first in order
to produce it. This model has found a correlation between proficiency level and the mode to
process vocabulary has been presented--lower bilinguals use the L1 to recognize vocabulary, and
proficiency bilinguals use the concept. For the present study, the RHM predicts that second
semester English monolingual students of Spanish will process cognates via their L1, given their
low proficiency level in Spanish. I would like to propose that degree of word similarity between
cognates is also a determining factor in addition to proficiency level. I predict that words with
greater similarities, such as similar cognates, will be related to the concept but words with fewer
similarities, such as partial cognates, will be related to the L1 word.
28
The second model pertinent to the present study is the BIA+ model. This model proposes
that when a proficient bilingual reads a letter string (a word), several lexical candidates—
regardless the language—are activated. Moreover, word recognition is affected not only by
cross-linguistic orthographic similarity effects, but also by cross-linguistic phonological and
semantic overlap, as well as the type of instruction or the language chosen for the task. Within
the present study, BIA+ model has been proposed to understand if phonology overlap occurs
while recognizing vocabulary. The model predicts that phonological overlap may induce
inhibition, that is, there is a negative influence of phonological similarities on word recognition.
Think-aloud protocols (TA) will be one method used to interpret students’ answers while they
are in the process of recognizing cognates (TA procedures are explained in more detail in
Chapter 3). The act of providing a TA protocol is seen by the experimenter as a factor that could
influence participants’ performance; however, no literature has been found to support this idea. If
the influence is positive, this could mean that think aloud can help students recognize and
process cognates. It is possible that the act of saying the cognates aloud could improve
comprehension of those words because it may help participants realize that the L2 word is
similar to the L1 word. However, as BIA+ model suggests, phonological interference can arise
and have a inhibit effect.
29
CHAPTER 3 METHODOLOGY
Participants
Ninety-one native-English speaking students enrolled in Elementary Spanish II at Auburn
University participated in the study; two of them were eliminated from the analysis because they
did not follow instructions during the experiment. The remaining 89 participants carried out the
study during one session in their eighth week of their second semester of Spanish, and they
received extra credit for their participation. Although and effort was made to have the same
number of participants in each group, the final groups resulted in unequal cell size: Control
Group had 22 participants, Group 1 had 28, Group 2 had 13, and Group 3 had 26. According to
the characteristics that students presented in the background questionnaire, the final division was
primarily based on whether or not participants had been exposed to Spanish before FLSP 1010,
that is, if they had studied Spanish before starting at Auburn University. Two groups were
created to make a distinction between participants who were exposed to Spanish. One group
included those who had studied Spanish during one or two years, and another included those who
had studied Spanish during three or four years. The other main division was based on whether or
not they had studied another language before. In general, most of the participants did not study
any language other than Spanish in high school. There was not a large difference between the
total number of participants who studied only one or two years, and the total number who studied
three or four years in high school. The characteristics of each group are shown in Table 3,
Appendix A. To avoid Hawthorne effects, all FLSP 1020 were asked to participate.
30
Experiment design and materials
As pointed out in Chapter 1 and 2, an explanation of what is considered a cognate and a
clarification of what is meant by “similarity” between words are needed. The following
definition and characteristics of a cognate were developed in this investigation: a cognate is
considered to be an item with a shared orthography, phonology, meaning and etymology in two
languages, in this case Spanish and English. In this work we do not consider borrowings as
cognates, (e.g., bar – bar; yogurt – yogur) due to the fact that borrowings are new words inserted
in the language lexicon, and they have not suffered the same historical evolution as cognates
introduced in the language centuries back. False friends are not included either in this study
because they do not share similar meaning and they are not the focus of the present research. In
addition, a proprietary scale of orthographic similarities between words was applied as follows:
1. Cognates that are orthographically identical are referred to as identical cognates
(“mural”–“mural”).
2. Cognates that differ by only one letter or when one letter in the English word is
replaced by two consecutive letters in its Spanish equivalent are referred to as similar
cognates (“inherent”–“inherente”; “civilization”–“civilización”; “family”-“familia”).
3. Cognates that differ by two non-consecutive letters or more than two non-consecutive
or consecutive letters are referred to as partial cognates: (“conclude”–“concluir”;
“access”-“acceder”).
These orthographic similarities or differences are counted from English to Spanish; an
orthographic difference includes the addition (“list”-“lista”), elimination (“blouse”-“blusa”) or
changes (“tomato”-“tomate”) of letters. Accent marks do not impact the classification, e.g.,
“utopia”–“utopia” are considered identical cognates. Phonological representations are also taken
31
into account. For example, the letters “ph” (as in “telephone”) are counted as only one letter for
the purposes of designating cognate status in the similarity scale. The pair “telephone”–
“teléfono” was considered a similar cognate because these words differ by two so-called letters:
the “ph” and the final “e”.
A minimal level of similarity has also been considered in the partial cognates. At least
three of the same letters have to remain. An example of a cognate that has not been classified as
partial for this study is that of “luxury”-“lujo”. These words share meaning and etymology (from
Latin, “luxus”), but do not share enough orthographic similarities to be readily identify as
cognates.
To solve the methodological issues mentioned in Chapter 2, cognates were given in a
context related to the material students regularly see in class. This fact is important because the
goal of this research is to contribute to the understanding of the acquisition process within the
classroom. With this in mind, the text that was used was an advertisement about three vacation
travel offers. The text developed for use within this experiment, follows the characteristics
(topic; grammar expressions and verb conjugations; vocabulary; and length–number of words)
that have been established for the students’ level (basic) by the National Council of State
Supervisors for Languages (NCSSFL). In other words, students could encounter a text like this
within their classroom environment. Another methodological issue that has been addressed in the
present study is the number of cognates that students found during the experiment. The number
of words was not a demanding cognitive effort for them related with their level: the total number
of cognates was 69 and 308 words for the total text. In terms of cognate frequency, each cognate
only appeared once in the text.
32
The cognates to be used in the three-advertisement texts were tested in a pilot before the
experiment was completed. The aim of the pilot was to find out if the cognates selected for the
main experiment were easy or difficult to recognize. To measure this, 208 students (divided into
15 classes) enrolled in Elementary Spanish I were tested. Not one of them was included in the
main study. To carry out the test, three word lists were created, one list for each “Travel offer
text”. Sixty percent of the words were cognates and 40 percent were distractors—non cognate
words. The total number of words distribution that appeared in each list is shown in Table 2,
Appendix A. A computer program was used to randomize the words lists:
http://www.random.org/lists/. Each list was presented to 5 classes, and students were asked to
translate every word that they could recognize from Spanish to English. Included with the list
was a background questionnaire to clarify previous experience of these FLSP 1010 students (all
documents are included in Appendix C, under Pilot documents). As a result, words that were
translated correctly by more than 40 percent of the students were considered very easy to
recognize, probably because they were words that they already knew for the classes; therefore
they were eliminated and replaced by other words in the main experiment.
In the full scale experiment, the final advertisement texts that participants were shown
contained a total of 308 words. Sixty-nine words (verbs, nouns and adjectives) were cognates: 21
were identical cognates, 23 were similar cognates, and 25 were partial cognates (see Table 1,
Appendix C, under Experiment documents).
Participants answered 17 comprehension questions in multiple-choice format. The
questions and answers were in English and tried to focus on the cognate vocabulary. The text
was divided into three parts or three travel advertisements (Sahara desert, Brazil and Egypt) and
a small conclusion. Participants read one part first, then answered 5 questions; next they read the
33
additional part, and then answer another 5 questions; the third text was presented afterwards
followed by another 5 questions; finally the small conclusion appeared, and after this part
participants answered 2 additional questions. The questions were presented in the order that the
information appeared in the text.
Participants were presented with the stimuli, the three vacation offer texts, using E-prime
2 Professional, and responded using both a keyboard and microphone. Two types of programs
were created to account for the two experimental conditions, one for the Control Group and
Group 1, and another for Group 2 and Group 3.
Procedure
All participants read an identical text and answered, in English, the comprehension
questions. Participants worked with the text differently depending on their group. All participants
had two hours to carry out the tasks, however, some groups used less time than others due to
their different experimental demands.
Groups
Participants were divided randomly into four groups: the control group (CG), Group 1
(G1), Group 2 (G2) and Group 3 (G3).
Control group: Participants silently read the text in a computer screen and answered the
comprehension questions. This is a typical activity that students would encounter in the
classroom and beginner textbooks, in an effort to represent what students usually do in the
classroom.
Group 1: Participants processed with a think-aloud (TA) protocol. They read the text in
the computer screen and answered the comprehension questions aloud verbalizing their thoughts
34
while being recorded as they completed the session (for more details about TA, see “Think-aloud
protocol” below).
Group 2: Participants had a different set of instructions. First, they saw the text word by
word in the computer screen. Reading the text word by word was intended a text-scanning
activity that many teachers use in their classrooms where students search for particular words
that they know or can recognize. They were asked to respond “yes” or “no” depending on
whether or not they could recognize the word that was presented on the screen. If they responded
“yes”, the next screen asked them for a written a translation of the word they saw before; if they
responded “no”, they skipped this step and moved to the next word. After reading the text word
by word, the complete text appeared and participants read it again and answered the
comprehension questions. This group was exposed twice to the cognates.
Group 3: Participants used a combination of the instruction set for Group 1 and Group 2.
They participated using a TA protocol and saw the text word by word in the computer screen,
translating the words that they recognized first. Afterwards they saw the whole text and
answered the comprehension questions.
The translation task that Group 2 and Group 3 carried out was focused on recognizing
and processing cognates, the main aim of this research. The TA protocol, used by Group 1 and
Group 3, was used in order to know what students were thinking while reading or answering the
questions, and these data has been used to analyze how students pronounce the cognate
vocabulary.
35
Think-aloud Protocol
In the think-aloud protocol, participants were asked to read the text presented in the
computer screen aloud and to verbalize their thoughts while reading, translating words, and
answering the comprehension questions. Each participant received special instructions and
practiced prior to beginning the study. The participants were given a three-sentence paragraph in
Spanish to read aloud and were asked to say whatever passed through their minds as they read.
Once training was completed, participants began the target reading task. Oral data was collected
using headsets and Audacity software.
Scoring Procedure
Variables: The dependent variables tested were participants’ 1) answers related to words
translated, and 2) responses to the comprehension questions. The independent variables tested
were 1) the type of cognates (identical, similar, and partial cognates), and 2) instructions given to
groups (CG, G1, G2, G3).
Two separate methods--quantitative and qualitative--were used to analyze the data. In
the quantitative scoring method, using Excel spreadsheet participants were awarded 1 point for
each correct answer and 0 points for each incorrect answering. In this method for the translation
task, spelling inaccuracies such as “casions” instead of “casinos”, “elixer” for “elixir”,
“instraments” for “instruments”, “collosel” or “colosal” for “colossal”, or “pharoes” for
“pharaohs”, were counted as correct answers because these misspellings can be attributed to the
difficulties of spelling within the English language. The number of correct answers for each
condition was compared. An analysis of variance (ANOVA) was carried out to determine if there
were significant differences between answers and groups. However, the present study is limited
by the small number of participants in G2 (13 participants) in compare to G3 (26 participants).
36
Qualitative analyses were applied to the incorrect answers. Two types of incorrect
answers were identified: Participants either incorrectly translated the word, or chose not to
respond at all. Among the incorrect translations, a pattern was noticed, so those semi-incorrect
answers have been taken into account as well. A semi-incorrect translation was defined as a
word in which a phonological pattern is detected in more than two participants when translating.
For example, some participants translated “occidental” as “accidental”, or “precio’ --“price”-- as
“precious”. TA protocols were used to support the analysis of the findings.
37
CHAPTER 4 RESULTS
The focus of the present study is cognate recognition and processing. In order to measure
the results, two types of data analyses were carried out: A quantitative analysis using an Excel
spreadsheet and a qualitative analysis. For the quantitative analysis two data sets were collected,
one from the responses to the reading comprehension task and the other one from the cognate
recognition task. The qualitative analysis was executed by utilizing the incorrect answers from
the cognate recognition task.
As mentioned before, in order to analyze the comprehension question results, all
responses were entered into an Excel spreadsheet. The total correct and incorrect answers
produced by each group and the corresponding percentages are shown in Table 4, Appendix A.
Results do not reflect a substantial difference between groups in terms of the comprehension
assessment: All the groups answered correctly approximately 65 percent of the comprehension
questions, and about 35 percent of the questions were answered incorrectly. An analysis of
variance (ANOVA) was carried out to determine if there were significant differences between
groups. It revealed that these differences were not significant with a P value of 0.50 (see Table 5,
Appendix A).
In particular to the analysis of the comprehension questions, the correct and incorrect
responses of each group (Control Group (GC), Group 1 (G1), Group 2 (G2) and Group 3 (G3))
were compared. Recall from Chapter 3 that in Groups 2 and 3 participants saw the text word by
word first, then read the whole text and answer questions while participants in the CG and G1
only read the whole text once before answered the questions. Results show that participants in
Groups 2 and 3 were slightly more successful at accurately responding to the comprehension
question task (achieving 67 and 68 percent of the total answers correct respectively) than CG and
G1 (achieving 66 and 63 percent of the total answers correct respectively). In fact, G1 had the
38
lowest number of correct answers: 63 percent; and G2 had the highest number of correct
answers: 68 percent, but these differences were not significant with a [P=0.22].
In addition, to explore the effects using a Think Aloud protocol might have had on the
comprehension, a statistical comparison between the responses of the Groups 2 and 3 (which
used TA) and Control Group and Group 1 (which did not use TA) was conducted. No significant
differences were found with a P value of 0.22. Interestingly, when groups that had the same text
conditions are compared, (CG vs. G1, or G2 vs. G3) groups that used TA (G1 and G3) were
slightly less successful at correctly answering the comprehension questions. However, these
differences were found not to be significant by an ANOVA [P=0.21].
The second data set derived from the present study pertains to cognate recognition. That
is, to the English translation provided by participants in Groups 2 and 3. Recall from Chapter 3
that in this study cognates were classified into three types: A) Cognates that are orthographically
identical are referred to as identical cognates (“mural”–“mural”). B) Cognates that only differ
by one letter or when one letter in the English word turns into two consecutive letters in its
Spanish equivalent are referred to as similar cognates (“inherent”–“inherente”; “family”-
“familia”). C) Cognates that differ by two non-consecutive letters or more than two non-
consecutive or consecutive letters are referred to as partial cognates: (“conclude”–“concluir”;
“access”-“accede”). To measure the differences between identical, similar, and partial
cognates, two different analyses were conducted. First, all the correct and incorrect translations
produced by participants in Groups 2 and 3 were compared. As mentioned in Chapter 3,
complete credit was also given to misspelled words, such as “elixer” for “elixir”, “excurcion” for
“excursion”, or “pecular” for “peculiar” which can be attributed to the difficulties of spelling
within the English language. As shown in Table 6, Appendix A, results suggested that words that
39
have more similarities, such as identical or similar cognates, were easier to recognize than
words that do not share as many scripts, such as partial cognates. Specifically, 68 percent of the
identical cognates were answered correctly; 55 percent of the similar; and 37 of the partial. In
order to test if these differences were significant an ANOVA test was carried out. Overall the test
reflects a significant difference for cognate degree of similarity with a P value < 0.001 (see Table
7, Appendix A). In addition, significant differences were found between identical and partial
cognates [P<0.001], and between similar and partial cognates [P<0.0001]. However, no
significant differences were found between identical and similar cognates [P=0.20].
Comparing groups 2 and 3 and conditions (G2 did not use a TA protocol and G3 used a
TA protocol), results indicate that G2 performed better on identical cognates [P=0.86], but there
were no significant differences between similar [P=0.38] and partial cognates [P=0.34] and
groups’ outcomes.
Proper nouns such as “Sahara”, “Tarzán”, “Aladín”, “Brasil”, etc., were inserted into the
text but to do the data analysis these proper nouns were not taken into account because as proper
names they refer to unique people or places. However, these proper names can be categorized as
identical, similar, or partial cognates, so an additional analysis focused on proper names was
carried out because it was explore that students did not perform well with these words regardless
of the word similarities. Results are shown in Table 8, Appendix- A.
The second type of data analysis conducted was qualitative, which focused on the
incorrect translation given by the participants. As presented in Chapter 3, incorrect responses
involved either the production of an incorrect translation of the Spanish word, or no response. In
terms of cognate processing, the revised hierarchical model (Kroll, & Cureley, 1988; Chen &
Leung, 1989; Kroll & Stewart, 1994) introduced in Chapter 1, was used to explore if students use
40
their L1 to process the visual word or if they use the concept. Notice that this model, although
proposed in the literature as a production model, has implications for the present study. In
particular, it predicts that novice students will rely on their L1 to process cognates. However,
results showed that participants used the concept to translate certain words, (see Table 9,
Appendix A). This has been observed because students did not use the English word while
translating, but the concept. Seventy-seven percent of the participants who translated the word
“Sahara” did it into “desert”; 25 percent translated “Aladín” into “Disney character”, and 9
percent translated “Tarzán” into “Disney character”, as well as 13 percent of the participants
who wrote a translation translated “anaconda” into “snake”.
Furthermore, some of the incorrect translations followed a pattern related to phonological
overlap, so those answers have been taken into account as well. A phonological overlap
translation was defined as a word in which a certain phonological pattern is identified in more
than two participants when translating. For instance, participants translated “occidental” as
“accidental”, “directos” as “directions” or “director”, “peculiar” as “movie”—“película” in
Spanish--, or “leyendas” (“legends” in English) as “they are reading”.
It has been interesting to note the translation of two words that could be classified as
phonological overlap translation but they do not fit in that category because they were produced
by only one participant instead of at least of two. An additional table with these two words has
been added (see Table 10, Appendix A).
Finally, the audio obtained from the TA protocols was used to clarify how students
pronounced the words, and what they were thinking while reading. As was detected within the
Questions section, neither the TA group nor the non-TA group produced substantially different
results while translating the target words.
41
CHAPTER 5 DISCUSSION
The research questions investigated in this study sought to explore cognate recognition
and processing in second language acquisition. First, the experiment tried to clarify whether or
not cognates can facilitate reading comprehension. Reading comprehension results do not reflect
a noticeable difference between groups nor do they reflect any kind of impact based on the
degree of cognate similarity—identical, similar, and partial cognate. In addition, results do not
demonstrate a significant influence based on the different conditions, e.g. whether using a Think
Aloud protocol while reading the text or not. A problem with the applied method may be the
main reason for the lack of differentiation between conditions and groups. It is possible that the
questions that were created for this experiment did not accurately assess cognate comprehension.
A better way to assess whether or not cognates can facilitate or inhibit reading comprehension
could be by using a different text for the Control Group with a reduced number of cognates and
compare it with a text similar to the one used in the present experiment.
Another topic addressed in the present study pertain to cognate recognition and cognate
degree of similarity. The following questions were explored: 1) Whether or not second semester
native English-speaking students of Spanish as an L2 recognized cognates in the written form; 2)
whether or not cognate degree of similarity had an effect on recognition; and 3) whether or not
these similarities could predict their performance. As it has been presented in the literature
review, Spanish textbook publishers and some researchers (Costa et al., 2000; Hall, 2002, 2009;
Holmes & Guerra, 1993; Hoshino & Kroll, 2008; Tercedor, 2010; Tonzar et al., 2009; Sherkina-
Lieber, 2004) have assumed that cognates are easy words to recognize. Nevertheless, results
obtained from this study provide support for the hypotheses discussed in Chapter 1 and 2. As
Lightbown et al. (1984), Tréville (1996), and Harley et al. (1986) proposed in their articles, the
42
results of the present investigation suggest that participants do not recognize cognates as easily
as presumed. First, students did not recognize all the cognates displayed in a text and, second, the
results indicate that there is a relationship between the hypothesis that students do not recognize
cognates easily and the orthographic similarity between the English/Spanish words, which I have
classified into three different categories: identical, similar, and partial cognates. A significant
difference was detected in the results between identical, similar and partial cognates. Identical
cognates were recognized significantly more often than similar and partial cognates; and
recognition of similar cognates was significantly higher than partial cognates. This shows that
the degree of similarity between cognates has an effect on novice-level Spanish students’
recognition and processing of vocabulary items. That being said, words that have more
similarities, such as identical or similar cognates, were significantly easier to recognize and
process correctly than words that do not share as many graphemes, such as partial cognates.
On the basis of these results, all cognates should not be considered as having the same
characteristics and impact on the learners. As mentioned before, researchers who worked with
cognates did not classify the characteristics or similarities between them. For example, Friel and
Kenninson (2002) and Sherkina–Lieber (2004) carried out a study using a 7 point scale task and
a 10 point scale task, respectively, to classify cognates’ similarities; however, they did not create
any classification based on the outcomes. As a result, what authors named as cognates were
actually a collection of words with very different characteristics and different ranks of difficulty.
Results from this study show that not all cognates are recognized and processed at the same
level. Consequently, we cannot consider cognates as easy words to recognize solely because they
share graphemes in common in both languages. Furthermore, even though identical cognates
had the highest number of correct translations, in many cases they were not translated correctly
43
by the participants with only a 68 percent of correct responses. Notice that a 68 percent correct
translation rate should not be taken as a substantial number of correct translations when the
target L2 word and the L1 word share the same spelling in both languages.
There are several factors that can influence the participants’ performance; one being their
familiarity with the target L1 word. The cognates chosen for this study are words with Latin and
Greek origins and some of these words are not frequently used in everyday English speech;
usually they are used in very formal settings. For example, none of the participants could
correctly translate the identical cognate “occidental”. “Occidental” is an English word, however,
English speakers generally use the term “Western”. Another identical cognate that presented
problems for participants was “irascible”; only one participant was able to translate it correctly.
“Afable” was also categorized as an identical cognate. Although the English and Spanish forms
differ by one letter (given that English “affable” has an additional “f”). The double grapheme
was counted as one sound [f] which causes the word to be classified as an identical cognate.
Only three students could translate it correctly. In common speaking, people do not use these
terms (generally speakers use “cranky”-“grumpy” for “irascible”, or “friendly” for “affable”).
Nevertheless, words such as “popular”, “agenda”, “cafeteria”, “ideal” and “anacondas”, were
correctly identified by almost 100 percent of participants who were able to give the correct
English equivalent. These words are more common in regular and everyday conversation.
Therefore, familiarity and similarity with the L1 word are both important in cognate recognition.
Due to this fact, it is also important to be aware of how different cognates can appear between
languages and how frequently they are used in everyday speech.
As in the comprehension questions, there were not significant differences between groups
and conditions (i.e., group that did not use a Think Aloud protocol, and group that used a Think
44
Aloud protocol) in terms of cognate recognition. Group 2 and 3 performed similarly; however,
participants that used a TA protocol performed slightly worse than participants who did not use
this method. This is an interesting outcome because even when significant differences are absent,
TA protocol did not positively influence their performance as was originally thought.
Nevertheless, one reason for this outcome could be that the students did not properly follow the
instructions while using Think Aloud protocol.
In addition to the cognate recognition, the present study focused on cognate processing,
using a qualitative data analysis. Two processing models were used to explore this topic: The
Revised Hierarchical Model (RHM) and the Bilingual Interactive Activation plus model (BIA+).
First, in terms of the RHM, the findings of the present study seem to contradict the predictions of
this model presented early. RHM proposed by Kroll and Stewart (1994) suggests that novice
learners used their L1 to process L2 vocabulary words and in doing so, predicted that participants
in the present study would process cognates using their L1. However, certain words with more
similarities, such as “Sahara”, “Tarzán”, “Aladín”, and “anaconda”, were related with the
concept by some of the participants. These participants did not translate “Sahara” as “Sahara”,
they translated it as “desert”; “Tarzán” or “Aladín” as “Disney character”, and “anaconda” as
“snake.” That being said, several participants used the concept to process the word instead of
using their L1. These results support the hypothesis presented in Chapter 2 stating that degree of
similarity would play a role in cognate processing.
The second processing model used in the present study the BIA+ model, assumes that
word recognition is affected not only by cross-linguistic orthographic similarity effects, but also
by cross-linguistic phonological overlap. As Dijkstra’s et al. (1999) model predicted, results
from this study show cross-language effects of phonological overlap between words. This
45
phonological overlap seems to produce a negative influence on a participant’s ability to
recognize a word due to phonological similarities between words, both crosslinguistically and
within their L1. The audio obtained by the TA protocol, shows a strong English pronunciation
influence, probably because of participant’s novice level. These pronunciations are thought to
have had a particular effect: it seemed that participants looked sometimes for a word in their
L1that was similar to the pronounced word and stopped paying attention to the written word.
Support for this conclusion is found in some audio recordings in which participants said: “This
sounds like ____”; and yet, none of them said “this word looks like ______”. The analysis
showed that participants provided incorrect answers for certain words and the pronunciations of
these words sounded much more similar to the English pronunciation. For example, they
pronounced “accidental” for the Spanish word “occidental”, “directions” or “director” for
“directo”, or “naturalize” for “naturaleza”, all of which are incorrect. In addition, the TA data
analysis revealed that in many cases participants looked for a Spanish word with similar spelling
and/or sound. For instance, while reading “leyendas” (“legends” in English) they interpreted it as
“leyendo” (“they are reading”), for “peculiar” they pronounced “película” (“movie” in English)
or for “area” they pronounced “arena” (“sand” in English). This phonological interference
while reading a word has been taken into account by Dijkstra’s et al. (1999), but it was not
mentioned in the aforementioned studies exposed in Chapter 1. However, it provides a very
interesting potential research area.
Some interesting observations were made after listening to the audio recordings created
by the participants who used a TA protocol. First, most of them said several times: “I don’t know
how to spell this, but I mean _____”. This supports the decision to give complete credit to
misspelled words, such as “elixer” for “elixir”, which can be attributed to orthographic
46
difficulties within the English language. Another important consideration was that there were
considerable differences between students while reading and processing a word. Some
participants took their time, tried to make connections with external things or activities, or talked
about how the word sounded. For example, with the word Occidental one participant said:
“occidental, occidental…that sounds like an accident, accidental, occidental…I’m not really
sure.” However, other students just read the word and said: “I think this means _____” or “I
don’t know this word.” The important point to take from these findings is that some students
look for similarities in order to recognize a word, and what the audio reveals is that they often
look for these relationships based on the pronunciation. Also, as Leow (2011) confirms, there are
different levels of awareness and diverse types of attention that can also affect L2 recognition
and comprehension.
Another finding that was not predicted in the hypotheses, but drew my attention, was the
relation to the suffix and correct/incorrect translations. A pattern within the correct answers was
found. Most of the words that have the suffix “–ion” in Spanish, produced a high percentage of
correct translations. For example, “dimensión”, “civilización”, and “excursión” were translated
more accurately. However, words with an ending of “–able”, such as “afable” or “formidable”,
and words ending with “–eza” such as “naturaleza” resulted in very low scores. These
observations can be related with the aforementioned ideas that there are some words that are
more common in every day speech, so their frequency is higher. Also, perhaps some suffixes are
easier to recognize than others.
47
CHAPTER 6 CONCLUSION AND FUTURE RESEARCH
The presented study tried to recreate as closely as possible a classroom environment
activity. A “travel offer” text which corresponded with the participants’ language level was given
to them to first read, and then they answered several comprehension questions. Reading the text
word by word was an attempt to replicate a text-scanning activity that many teachers use in their
classrooms where students search for particular words that they know or that they can recognize.
The results of this study support the findings of Lightbown et al. (1984), Tréville (1996), and
Harley et al. (1986) which concluded that students do not recognize cognates and, due to this,
instructors should teach students how to recognize and work with cognates. As well, to believe
that a cognate, only because of having a common root and shared graphics, is an easy word to
recognize is to negate that there are multiple differences amongst cognates, and cognate
characteristics. An important aspect of this study is the degree of similarities scale between
cognates—identical, similar, and partial—and their effects on word recognition.
In terms of phonological overlap, the presented ideas have the potential of creating a new
and interesting strand of research. Results show that phonology plays an important role while
recognizing a written word.
This study could be reconfigured with comprehension questions that are more targeted to
elicit a cognate response in an effort to clarify whether or not cognates aid in reading
comprehension. Also, the present study can be modified using a Control Group with the same
text as the other groups but with fewer cognates would be an interesting alternate technique to
measure if cognates actually help or not with reading comprehension. The present study also is
limited by the small number of participants in G2 (13 participants) in compare to G3 (26
participants). These data do not meet the assumptions of the ANOVA model. That is, the test is
48
less powerful. Future studies may equal the number of participants per group as much as they
can to obtain ideal conditions.
For further research, it would be interesting to study the frequency of words in speech
and students’ familiarity with the target words. As mentioned before, not all cognates are easy
words even when they are completely equal orthographically. One reason that has been
considered as a cause of these results is that most of the presented vocabulary is not well-known
terminology. As well as, it can be interesting to research whether or not the function of the word
(nouns, verbs, adverbs, etc.) and the kind of prefix or suffix have an influence on cognate
recognition. In addition, the assumption that students read a word and pay attention to the written
form appears to be a weak point in this investigation and would benefit from further research.
The most important conclusion from this experiment is that target word recognition was
significantly influenced by the type of cognate—whether the words were identical, similar, or
partial cognates. Likewise, given the pedagogical implications, teachers should not assume that
just because cognates are used will students recognize them. It is important to include more
activities focused on word recognition in text-books or language curriculums to train students to
identify cognates and use them appropriately.
49
References
Chen, H. (1990). Lexical processing in a non-native language: effects of language proficiency
and learning strategy. Journal of Experimental Memory, and Cognition, 18(3), 279-288.
Chen, H. & Leung, Y. (1989). Patterns of lexical processing in a nonnative language. Journal of
experimental memory, and cognition, 15(2), 316-325.
Chomsky. (1959). On certain formal properties of grammars. Information and control, 2(2), 137-
167.
Costa et al. (2000). The cognate facilitation effect: implications for model of lexical access.
Journal of experimental psychology: learning, memory, and cognition, 26(5), 1283-1296.
Dijsktra, T. & Van Hueven, Walter J.B. (1998). Orthographic neighborhood effects in bilingual
word recognition. Journal of memory and language. 39, 458-483.
Dijsktra, T. et al. (1998). Interlingual homograph recognition: Effects of task demands and
language intermixing. Bilingualism: Language and Cognition, 1, 51-66.
Dijsktra, et al. (1999). Recognition on cognates and interlingual homographs: The neglected role
of phonology. Journal of memory and language. 41, 496-518.
Dijsktra, T. & Van Hueven, Walter J.B. (2002). The architecture of the bilingual word
recognition system: From identification to decision. Bilingualism: Language and
Cognition, 5(3), 175-197.
Dijsktra, T. & Van Hueven, Walter J.B. (2002). Modeling Bilingual word recognition: Past,
present and future. Bilingualism: Language and Cognition, 5(3), 219-224.
Di Pietro, R.J. (1971). Language Structures in Contrast. New York, NY: Newbury House.
De Groot et al. (1994). Forward and backward word translation by bilinguals. Journal of memory
and language, 33, 600-629.
50
Dulay, H. & Burt, M. (1973). Should we teach children syntax? Language Learning, 23, 95-123.
Ellis, R. (1986). Understanding Second Language Acquisition. Oxford: Oxford University Press.
Fisiak, Jacek. (1980). Theoretical issues in contrastive linguistics. Amsterdam: Amsterdam
studies in the theory and history of linguistic science.
Friel, B. & Kennison, S. (2001). Identifying German – English cognate, false cognates, and non-
cognates: methodological issues and descriptive norms. Language and Cognition, 4(3)
249-274.
Gass, S. (1990). Second language vocabulary acquisition. Annual review of applied linguistics 9,
92-106.
González Barrera, M. (2010). Predicción y solución de dificultades que un alumno de habla
inglesa pueda tener al aprender fonética española. Biblioteca virtual redELE, 11.
Retrieved from:
http://www.educacion.es/redele/Biblioteca2010/MaikaGonzalez/Memoria.pdf
Hall, C. (2002). The automatic cognate form assumption: evidence for the parasitic model of
vocabulary development. International review of Applied Linguistics, 40(1), 69-19.
Hall, C. et al. (2009). Learners’ implicit assumptions about syntactic frames in new L3 words:
the role of cognates, typological proximity, and L2 status. Language Learning, 59(1),
153-202.
Harley, B. et al. (1986). The effects of early bilingual schooling on first language skills. Applied
Psycholinguistics, 7, 295-322.
Holmes, J. & Guerra Ramos, R. (1993). False friends and reckless guessers: observing cognate
recognition strategies. In Thomas Huckin, Margot Haynes & James Coady. Second
Language Reading and Vocabulary Learning (86-108). Norwood, NJ: Ablex.
51
Hoshino, N. & Kroll, J. (2008). Cognate effects in picture naming: does cross-language
activation survive a change of script. Cognition, 106, 501-511.
Kroll, J. F. & Curley, J. (1988). Lexical memory in novice bilinguals: the role of concepts in
retrieving second language words. In M. M. Gruneberg, P. E. Morris & R. N. Sykes.
Practical aspects of memory: current research and issues, volume 2. London: Wiley.
Kroll, J. F. & Stewart, E. (1994). Category interference in translation and picture naming:
evidence for asymmetric connections between bilingual memory representation. Journal
of Memory and Language, 33, 149-174.
Kroll, J. F. et al. (2002). The development of lexical fluency in a second language. Second
Language Research, 18(2), 137-171.
Kroll, J. F. et al. (2010). The Revised Hierarchical Model: A critical review and assessment.
National Institute of Health, Public Access. Retrieved from:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2910435/
Lado, R. (1957). Linguistics across cultures: Applied linguistics for language teachers.
University of Michigan Press: Ann Arbor.
Lado, R. (1965). Pattern of difficulty in vocabulary. Teaching English as a second language.
New York: McGraw Hill Book Co.
Levenston, E.A. (1979). Second language acquisition: issues and problems. Interlanguage
Studies Bulletin 4, 147-160.
Leow, R. P. (2011). A study of the role of awareness in foreign language behavior. Studies in
second language acquisition, 22, 557-584.
52
Lightbown, P. & Libben, G. (1984). The recognition and use of cognates by L2 learners. In
Second Language a cross-linguistics perspective (393-417). New York, NY: Newbury
House.
Lobo, Felix. (1966). A 10,000 Spanish word vocabulary expanded from 3,000 English cognates.
(Doctoral dissertation). Retrieved from Georgetown University, Washington D.C.
Moreno Fernandez, F. (2000). Ejercicios de fonética española para hablantes de inglés. Madrid:
Arco/Libros.
Oxford, R. & Scarcella, R. (1994). Second language vocabulary learning among adults: state of
the art in vocabulary instruction. System, 22(2), 231-243.
Politzer, R. (1978). Some reflections on the role of linguistics in the preparation of bilingual
teachers. Proceedings from American Council on Teaching of Foreign Languages. San
Francisco, CA.
Potter, M. et al. (1984). Lexical and conceptual representation in beginning and proficient
bilinguals. Journal of Verbal Learning & Verbal Behavior, 23, 23-38.
Serradilla Castaño, A. (2000). La enseñanza de la pronunciación en el aula: Una experiencia con
estudiantes anglohablantes. In M. Franco, C., Soler, J. de Cos, M. Rivas, & F. Ruiz.
Nuevas perspectivas en la enseñanza del español como lengua extranjeras. Proceedings
from X Congreso de ASELE. Cádiz, Spain.
Sherkina-Lieber, M. (2004). The cognate facilitation effect in bilingual speech processing; the
case of Russian-English bilingualism. Cahiers linguistics d’Ottawa., 32, 108-121.
Sunderman, G. & Kroll, J. F. (2006). First language activation during second language lexical
processing. Studies in Second Language Acquisition, 28, 387-422.
53
Tercedor, M. (2010). Cognates as lexical choices in translation: interference in space-constrained
environments. Target, 22(2), 177-193.
Terrel, T. (1989). Teaching Spanish pronunciation in a communicative approach. In Bjarkman,
P. & Hammond, R. American Spanish Pronunciation Theoretical and Applied
Perspectives. (196-214). Washington: Georgetown University Press.
Tonzar, C. et al. (2009). L2 vocabulary acquisition in children: effects of learning method and
cognate status. Language Learning, 59(3), 623-646.
Tréville, M. (1996). Lexical learning and reading in L2 at the beginner Level: the advantage of
cognates. The Canadian Modern Language Review, 53, 173-190.
Zanón, F. (2007). Psicolingüística y didáctica de las lenguas: una aproximación histórica y
conceptual. Marco-ELE. Revista-Didáctica 5. Retrieved from:
http://www.marcoele.com/descargas/5/zanon-psicolinguistica.pdf
54
APPENDIX A
Tables
Table 1. Steeps to cover in the different models (Source: Adapted from Potter et el., 1984)
Word-association Concept-mediation
Translation Picture-naming Translation Picture-naming
1. Recognize the word
2. Associate L1 word with L2
3. Name the word
1. Recognize the picture
2. Activate the concept
3. Associate L1 concept to L1 word
4. Associate L1 word with L2 word
5. Naming the word
1. Recognize the word
2. Activate the concept
3. Associate the L2 word
4. Name the word
1. Recognize the picture
2. Activate the concept
3. Associate the L2 word
4. Name the word
55
Table 2 Distribution of cognates in pilot lists
Identical Similar Partial Total cognates Non-cognates
List 1 7 7 8 22 14
List 2 7 8 8 23 14
List 3 7 8 8 23 14
56
Table 3. Characteristics of participants in each group
No Spanish
in High
School
No Spanish
in HS, but
studied other
language
Studied
Spanish
previously for
1-2 years
Studied
Spanish
previously for
1-2 years, and
also other
language
Studied
Spanish
previously
for 3-4 years
Spanish
previously
for 3-4
years, and
also other
language
Total
number
of
students
CG 1 1 7 1 11 1 22
G1 1 2 12 2 10 1 28
G2 2 3 3 5 13
G3 2 11 5 8 26
2 7 33 9 34 2
57
Table 4. Percentages of correct and incorrect answers per group
Type of
group
Correct
Answer
Percentage Incorrect
Answer
Percentage Total
Questions
CG 248 66% 126 34% 374
G1 302 63% 174 37% 476
G2 151 68% 70 32% 221
G3 298 67% 144 33% 442
0
20
40
60
80
100
ControlGroup
Group 1 Group 2 Group 3
Correct Answers
Incorrect answers
58
Table 5. Analysis of variance (ANOVA) of correct and incorrect answers
Source of
Variation
df MS F P-value F crit
Between Groups 3 2.96307812 0.790485528 0.5025119 2.7119214
Within Groups 85 3.748428043
Total 88
0
2
4
6
8
10
12
14
16
1 2 3 4
Series1
CG G1 G2 G3
59
Table 6. Correct and incorrect answers based on type of cognates
Identical
Correct Incorrect
Similar
Correct Incorrect
Partial
Correct Incorrect
G2 67 33
58 42
37 63
G3 68 32
58 42
37 63
Correct Incorrect Correct Incorrect Correct Incorrect ___________________ ____________________ __________________ Identical Similar Partial
0
20
40
60
80
100
Group 2 Group 3
60
Table 7. Analysis of variance (ANOVA) of correct and incorrect answers based on type of
cognates
Source of Variation df MS F P-value F crit
Between Groups 2 327.8717949 27.183046 2.224E-10 3.0758526
Within Groups 114 12.06162843
Total 116
0
2
4
6
8
10
12
14
16
18
20
1 2 3
Series1
Identical Similar Partial
61
Table 8. Proper nouns percentages
_________________________________________________________________________________
Identical Similar Partial
Sahara Aladín Tarzán Cairo Sinai Bereber Brasil Egipto
Group 2 46 46 92 77 31 0 100 92
Group 3 50 46 88 77 27 0 96 69
0
20
40
60
80
100
Group 2
Group 3
62
Table 9. Concept used to translate certain words
Spanish Word Translation Number of
translations
Total number of
tranlations
Percentage
Sahara Desert 10 13 76%
Aladín Disney
character
3 12 25%
Anaconda Snake 3 23 13%
Tarzán Disney
character
2 22 9%
63
Table 10. Words that showed phonological overlap
A) Primary phonological overlap words
B) Additional phonological overlap words table
Área Area /ˈɛəriə/ Sand 3%
Estresado Stressed /strɛsɪd/ interested /ˈɪntərəstɪd/ 3%
Spanish
word
Correct
translation
Pronunciation Phonological
overlap
Pronunciation Percentage
Occidental Occidental /ˌɒksɪˈdɛntl/ Accidental /ˌæksɪˈdɛntl/ 28%
Irascible Irascible /ɪˈræsəbəl/ Irresistible /ˌɪrɪˈzɪstəbəl/ 5%
Formidable
Formidable /ˈfɔrmɪdəbəl/ Formulate –
formable
/ˈfɔrmyəˌleɪt/ /fɔrməbəl/
5%
Peculiar Peculiar /pɪˈkyulyər/ Movie 13%
Atípica Atypical /eɪˈtɪpɪkəl/ Anticipate /ænˈtɪsəˌpeɪt/ 10%
Gastronomía Gastronomic /ˌgæstrəˈnɒmɪk/ Astronomical /ˌæstrəˈnɒmɪkəl/ 8%
Subacuático Subaquatic /ˌsʌbəˈkwætɪk/ Suburban /səˈbɜrbən/ 5%
Directos
Direct /daɪrɛkt/ Director-
direction
/daɪˈrɛktər/ / ˈdaɪrɛkʃən/
41%
Descubrir Discover /dɪˈskʌvər/ Describe /dɪˈskraɪb/ 23%
Leyendas
Legends /ˈlɛdʒənd/ They are
reading
23%
Naturaleza Nature /ˈneɪtʃər/ Naturalize /ˈnætʃərəˌlaɪz/ 46%
Sarcófagos Sarcophagus /sɑrˈkɒfəgəs/ Sacrifices /ˈsækrəˌfaɪs/ 10%
Precio Price /praɪs/ precious /ˈprɛʃəs/ 31%
64
APPENDIX B
Figures
Figure 1. The three theoretical models (Source: Adapted from De Groot et. al, 1994)
Figure 1
Figure 2
Figure 3
1. Word association model
2. Concept association
mediation model
3. Intermediate model
65
Figure 2. Bilingual interaction model (Source: Adapted from Dijkstra et al. 1998)
Language Word level Letter level Feature level
Visual input
66
Figure 3. The BIA+ model. (Source: Adapted from Dijstra & Van Heuven, 2002)
Visual input
Visual input
Identification system
Language nodes Semantics
Lexical orthography Lexical
phonology
Sublexical orthography Sublexical
phonology
L1/L2
Task Schema Specifies processing steps at hand
Receives continuous input from the identification
system
Decision criteria determine when a response is made
based on relevant codes
___________
___________
___________
___________
___________
___________
______
67
APPENDIX C
Experiment documents
Table 1 Cognates that were presented in the study (proper nouns are in italics).
Identical Similar Partial
Text 1:
Sahara
Occidental
Aladín - Aladdin
Dimensión - dimension
Afable - affable
Popular
Oasis
Regional
Àrea - area
Text 2:
Irascible
Ideal
Tarzán - Tarzan
Anacondas
Cafeterías – cafeterias
Agenda
Cruel
Colosal – colossal
Text 3:
Elixir
Cleopatra
Sublime
Cairo
Formidable
Casinos
Sinaí - Sinai
Coral
Introduce
Peculiar
Madrid
Casablanca
Text 1:
Bereber - berber
Civilización - civilization
Atípica – atypical
Rica - rich
Incluye –include
Excursiones- excursions
Instrumentos – instruments
Dunas -dunes
Text 2:
Brasil - Brazil
Tragedia – tragedy
Evento – event
Intrépido – intrepid
Virgen – virgin
Jungla – jungle
Tabernas – taverns
Opulentas – opulent
Text 3:
Seducir – seduce
Melodías - melodies
Visita - visit
Gastronomía – gastronomy
Subacuático - subaquatic
Pirámide – pyramid
Explora - explore
Directos - direct
Excepto – except
Text 1
Descubrir – discover
Leyendas - legends
Desierto – desert
Único - unique
Palmeras –palm trees
Dátiles –dates
Paraíso -paradise
Text 2:
Estresado – stressed
Desconectar – disconnect
Contemplar - contemplate
Naturaleza – nature
Tucanes - toucans
Cocodrilo – crocodile
Accede – Access
Text 3:
Egipto – Egypt
Faraones – pharaoh
Invitamos - invite
Conquistar – conquer
Culmina - culminate
Admirando – admiring
Momias - mummies
Sarcófagos - sarcophagus
Precio – Price
Ofertas - offers
Aeropuerto – airport
68
Easy Viajar (115 Magnolia Avenue, Auburn, AL).
Texto 1.
Viaje 1. El Sahara Occidental.
Si quieres descubrir el mundo de Aladín, escuchar las leyendas de los bereberes1 y ver otra
dimensión, este viaje al desierto es para ti. Conoce una única y vieja civilización con gente
atípica, pero también muy afable.
El viaje incluye varias excursiones; la más popular es al oasis Tangounite con altas palmeras.
Aquí puedes comer la rica comida, como el tajine de pollo y dátiles, o comprar instrumentos de
artesanía regional. Viaje también a las dunas de Erg Chegaga, en el área de AbhDakla. ¡Todo un
paraíso!
Texto 2.
Viaje 2. La selva2 de Brasil.
¿Te sientes estresado? ¿Estás irascible? Si buscas desconectar del trabajo y de la tragedia diaria,
éste es el viaje ideal. No lo pienses más y ven a vivir el evento de tu vida. ¡Siéntete como
Tarzán! Disfruta de un viaje intrépido a la selva virgen donde puedes contemplar la naturaleza:
tucanes, loros, anacondas, cocodrilos, etc.
En este viaje vas a acceder a la jungla, así que olvídate de las tabernas, cafeterías y casas
opulentas. Siete días para olvidar la agenda, los profesores o el jefe cruel. Un viaje colosal.
Texto 3.
Viaje 3. El elixir de la vida.
Egipto es el país de Cleopatra y los faraones. ¡Déjate seducir por el sublime paisaje, los sabores y
las melodías encantadoras! Visita la ciudad El Cairo, donde te invitamos a disfrutar de su
gastronomía, de su formidable gente y de sus casinos.
Bucea por el mundo subacuático del Sinaí y déjate conquistar por su arrecife3 de coral. Culmina
tu viaje admirando las pirámides de Keops, Kefren y Micerinos. Explora estos edificios e
introduce tu mente en un mundo peculiar de momias y sarcófagos.
Precio de las ofertas: Sahara – 1100 €; Brasil -800 €; Egipto – 1200 €.
Todos los vuelos son directos desde el aeropuerto de Madrid, excepto el vuelo al Sahara con
escala en Casablanca.
1 Desert tribe. 2 Rain forest. 3 Reef
69
English Text.
Easy Viajar (115 Magnolia Avenue, Auburn, AL).
Travel 1. The Sahara Occidental.
If you want to discover Aladdin’s world, listen to Berbers’ legends and see another dimension,
this trip to the dessert is for you. Get to know a new and unique civilization with atypical, but
very affable people. The trip includes several excursions; the most popular is to the Tangounite
oasis with tall palm trees. Here you can eat the delicious food, like tajine with chicken and dates,
or buy regional instruments. Another trip is to the Erg Chegaga dunes in AbhDakla area. A real
paradise!
Travel 2. The rainforest of Brazil.
¿Are you stressed? ¿Are you irascible? If you are looking to disconnect from work and daily
tragedy, this trip is ideal. Do not think about it anymore and come to experience the event of a
lifetime. Feel like Tarzan! Enjoy an intrepid trip to the virgin rainforest, where you can
contemplate the nature: toucans, parrots, anacondas, crocodiles, etc.
In this trip you are going to access to the jungle. Forget taverns, cafeterias, and opulent houses.
Seven days trip to forget the agenda, your teachers or a cruel boss. A colossal trip!!
Travel 3. The life elixir.
Egypt is the country of Cleopatra and pharaohs. Let yourself be seduced by the sublime
landscapes, its flavors and its charming melodies! Visit the city Cairo, where we invite you
enjoy its gastronomy, its formidable people, and casinos. Scuba-dive around the sub-aquatic
world of Sinai and let yourself be conquered by its coral reef. Culminate your trip admiring the
pyramids of Keops, Kefren and Micerinos. Explore these buildings and introduce your mind to a
peculiar world of mummies and sarcophagus’.
Travel offers’ prices: Sahara – 1100 €; Brazil -800 €; Egypt – 1200 €.
All flights are direct flights from Madrid airport, except the flight to Sahara that stops in
Casablanca.
70
Questions:
1. El Sahara Occidental.
1. What can you enjoy from the Bereberes?
a. Their dances and music
b. Their flying carpets
c. Their legends
2. What adjectives better describe the people of the Sahara?
a. Distinct and friendly
b. All of them are family
c. Easy going and just like Westerners
3. Which is the most famous tour of the Sahara?
a. The oasis of Tangounite
b. The dunes of Erg Chegaga
c. None of the above
4. Which are the ingredients in tajine?
a. Chicken and grapes
b. Chicken and dates
c. Chicken and plums
5. What can you buy from the Bereberes?
a. Musical instruments
b. Bracelets and necklaces
c. Local handcrafts
2. La selva de Brasil
1. What is the goal of this trip?
a. To meet new people
b. To disconnect from family and daily drama
c. To enjoy wild animals
2. What type of trip will this be?
a. It will be a relaxing journey
b. It will be an exciting trip
c. It will be a boring trip
3. What are you going to be able to see?
a. Nature and animals
b. Skyscrapers and business men
c. Resorts and beaches
71
4. What are you NOT going to find in the jungle?
a. Luxury homes
b. Crocodiles, iguanas, parrots, toucans, etc.
c. Nice weather
5. What will this trip make you forget?
a. The trip schedule
b. Your back pack, the guide will give you one
c. Your horrible bosses and teachers
6. El elixir de la vida
1. What things will entice you?
a. The landscape, flavors and charming melodies
b. The handsome and pretty Egyptians
c. Fast-paced taxi drivers
2. What did the Cairo advertisement say you would enjoy?
a. Food and different smells
b. Drinking liqueurs
c. Scuba diving
3. What things will you be able to encounter?
a. New species of fish
b. The foundations of undiscovered pyramids
c. Ancient buildings and mummies
4. At what point of the trip will you see the pyramids?
a. At the beginning
b. At the middle
c. At the end
5. How is the new world that you can discover?
a. It is boring
b. It is unusual
c. It is magic
5. General questions
1. Which flight is NOT a direct flight? Where does it stop?
a. All of them are direct flights
b. The flight to Sahara is not direct; it stops in Casablanca.
c. The flight to Brazil is not direct; it stops in Madrid
2. Which trip is the least comfortable?
a. Sahara
b. Brazil
c. Egypt
72
Pilot documents
Background Questionnaire
Answer every question. If a question does not apply to you, write “N/A”.
1. Age______________________________________________________________
2. Class year: Freshman / Sophomore / Junior / Senior
3. Major____________________________________________________________
4. Minor____________________________________________________________
5. What do you consider your native language(s) to be? _____________________
6. How old were you when you were first exposed to Spanish? ________________
7. On a scale from one to seven, seven being excellent, please select your level of proficiency in
terms of understanding spoken Spanish by a native Spanish speaker.
1 2 3 4 5 6 7
8. High school Spanish (circle all that apply or add to the category other if applicable. If you did
NOT take Spanish classes while in high school, circle N/A here):
Spanish I Spanish III Spanish V
Spanish II Spanish IV AP Spanish/IB Spanish
Other: _____________________________________________________
9. College Spanish (circle all the Spanish courses, indicating whether they were taken on campus or
abroad. Add to the category other if applicable):
Elementary I (on campus / abroad)
Elementary II (on campus / abroad)
Intermediate I (on campus / abroad)
Other _______________________
10. On a scale from one to seven, seven being very much, please select how much the following
factors contributed to you learning Spanish:
1 2 3 4 5 6 7
Interacting with friends: ______________ Interacting with family: _________________________
Reading: ___________________________ Taking language courses: ________________________
Language tapes / self-instruction: _______ Watching TV: _________________________________
Listening to the radio: ________________ Being immersed in a Spanish-speaking country: _____
73
11. Travel to Spanish-speaking countries:
Country:
Length of
time:
Reason (e.g. vacation,
Spanish language course,
dad worked there):
Use of Spanish while in the
country:
12. Other languages studied or spoken:
Language:
Length of time studied or
spoken:
Use of the language (in what ways
do you or did you used to engage in
the language?):
13. Other language experience not mentioned above:
14. Did you study abroad last Summer 2011? Yes / No
15. If so, what location? Costa Rica / Salamanca / Madrid
74
Text.1. Read the list of words. Circle the ones that you can recognize, and give its English
translation.
1. Cartera _________________________
2. Lápiz____________________________
3. Palmeras_________________________
4. Viaje____________________________
5. Oferta___________________________
6. Popular__________________________
7. Oasis____________________________
8. Descubrir________________________
9. Diferente_________________________
10. Incluye__________________________
11. Rica_____________________________
12. Desierto_________________________
13. Tapa____________________________
14. Gente___________________________
15. Dunas___________________________
16. Región__________________________
17. Local____________________________
18. Sartén_________________________
19. Productos_________________________
20. Familiar__________________________
21. Paraíso___________________________
22. Único____________________________
23. Artesanía_________________________
24. Nueva____________________________
25. Mesa_____________________________
26. Mundo___________________________
27. Dátiles____________________________
28. Quieres___________________________
29. Cultura___________________________
30. Repollo___________________________
31. Comer____________________________
32. Leyendas__________________________
33. Dimensión_________________________
34. Excursiones_______________________
35. Escuchar__________________________
36. Occidental________________________
75
Text.2. Read the list of words. Circle the ones that you can recognize, and give its English
translation.
1. Verde____________________________
2. Ven_____________________________
3. Iguana___________________________
4. Drama___________________________
5. Desconectar______________________
6. Estresado________________________
7. Vivir____________________________
8. Tropical_________________________
9. Cocodrilo________________________
10. Cafeterías________________________
11. Irritable_________________________
12. Observar________________________
13. Jungla___________________________
14. Lujo_____________________________
15. Casas____________________________
16. Siete_____________________________
17. Temperatura______________________
18. Tucanes__________________________
19. Loro_____________________________
20. Pienses___________________________
21. Día______________________________
22. Aventura_________________________
23. Accede___________________________
24. Directo___________________________
25. Moderno_________________________
26. Disfruta__________________________
27. Trabajo__________________________
28. Ideal_____________________________
29. Cesta____________________________
30. Memorable_______________________
31. Restaurante______________________
32. Selva____________________________
33. Naturaleza_______________________
34. Vida____________________________
35. Precio___________________________
36. Extremo_________________________
37. Vela_____________________________
76
Text.3. Read the list of words. Circle the ones that you can recognize, and give its English
translation.
1. Natural_____________________________
2. Guapo______________________________
3. Excepto_____________________________
4. Recomendamos______________________
5. Aroma______________________________
6. Seducir_____________________________
7. Conquistar__________________________
8. Sarcófagos__________________________
9. Bucea______________________________
10. Vista_______________________________
11. Mundo_____________________________
12. Paisaje______________________________
13. Faraones____________________________
14. Pirámides___________________________
15. Momias_____________________________
16. Casinos_____________________________
17. Sientes______________________________
18. Viaje_______________________________
19. Música_____________________________
20. Bolígrafo____________________________
21. Arrecife____________________________
22. Peculiar_____________________________
23. Color_______________________________
24. Sabor_______________________________
25. Gastronomía_________________________
26. Coral_______________________________
27. Mente_______________________________
28. Encantador__________________________
29. Explora_____________________________
30. Culmina_____________________________
31. País_________________________________
32. Admirando__________________________
33. Introduce____________________________
34. Submarino____________________________
35. Déjate________________________________
36. Gente________________________________
37. Aeropuerto___________________________